Add native multilayer node-embedding support (state + physical targets) via existing py3plex embedding seams

[Copilot]

This PR adds native multilayer embedding support built around py3plex’s (node, layer) state-node model while preserving existing embedding seams and backward compatibility.

What’s included

• Deterministic state-node indexing and round-trip mapping via NodeLayerIndexer.
• Shared multilayer embedding foundation (MultiLayerEmbeddingConfig, BaseMultiLayerEmbedding) with:
  • target: state | node | both
  • state→node reduction: mean | sum | max (with guarded attention)
  • interlayer/coupling controls and metadata.
• New multilayer-capable embedding models integrated into existing APIs:
  • SupraNode2VecEmbedding
  • SupraSpectralEmbedding
  • SupraNetMFEmbedding
  • MNEEmbedding
  • MELLEmbedding
  • MultiLayerGNNEmbedding (explicit experimental scaffold).
• multi_layer_network.embed(...) dispatch extended for:
  • supra_node2vec, supra_spectral, supra_netmf, mne, mell, multilayer_gnn
  • while preserving existing aliases (multiplex_node2vec, supra_adjacency, layer_regularized).
• Embedding evaluation utilities extended with edge operators and embedding-native link prediction / richer node classification metrics.

Feedback follow-ups completed

• Docs were aligned for the new embedding surface:
  • API index now documents additional embedding modules (multiplex, evaluation, similarity).
  • User guide includes a dedicated “Native Multilayer Embedding API” section with usage examples and supported methods.
  • Docs tests were extended to enforce this coverage.
• Added missing property-based tests for multilayer embedding invariants in:
  • tests/property/test_multilayer_embedding_properties.py
  • covering indexer determinism + round-trip behavior, reducer correctness (mean/sum/max), and guarded attention behavior.

Validation

• Targeted docs and embedding tests were run successfully, including the new property-based test module and multilayer embedding regression tests.

[SkBlaz]

@copilot make sure docs are aligned

[Copilot]

@copilot make sure docs are aligned

Done — docs are now aligned with the new multilayer embedding APIs in both the API index and embedding user guide, and docs tests were updated to enforce this coverage. Addressed in commit 6f202ca (plus the main docs alignment commit 2918feb).

[SkBlaz]

@copilot you are missing property based tests

[Copilot]

@copilot you are missing property based tests

Added property-based coverage for the multilayer embedding work in tests/property/test_multilayer_embedding_properties.py (indexer determinism/round-trip, reducer invariants, and guarded attention reducer behavior). Addressed in commit aa5b68c.